Difference between revisions of "Talk:GRASS Community Sprint Prague 2012"

From GRASS-Wiki
Jump to: navigation, search
m (Update the GDAL-GRASS interface to GRASS 7)
(Signal handling)
Line 86: Line 86:
 
=== Signal handling ===
 
=== Signal handling ===
  
The Python based scripts should catch CTRL-C to avoid that clutter is left behind and that ugly traceback is printed into the terminal.
+
* The Python based scripts should catch CTRL-C to avoid that clutter is left behind and that ugly traceback is printed into the terminal.
  
 
Suggestion by Anne Ghisla (yet not working... ):
 
Suggestion by Anne Ghisla (yet not working... ):
Line 109: Line 109:
 
</source>
 
</source>
  
 +
 +
* Currently if vector creation modules fail to create/import they still leave behind the empty skeleton of a map, so you need to use --o for the next time. It would be nice if cleanup could occur. This doesn't happen (visibly) with raster maps as they are composed in $MAPSET/.tmp/ and only moved into the main $MAPSET elements once everything is finished. Vector maps on the otherhand write to the MAPSET in-place.
  
 
=== Fix wxGUI problems ===
 
=== Fix wxGUI problems ===

Revision as of 12:53, 22 May 2012

Wishes for the Community Sprint

  • Visual Guide to GRASS-GIS modules: scratch a collection of per-module dedicated, simple, minimalistic yet attractive, bullet-proof visual examples
    • setup specific styling rules (?)
    • end-up with some independent and beautiful TeXy PDF
    • visually enrich existing manuals and/or grass-wiki
  • Review of the python + ctypes example scripts.
The examples given in the code for doc/python/raster_example_ctypes.py, vector_example_ctypes.py and on the Ctypes wiki page have so far for me all been broken in some way or another. Either grass7 API in grass6 example, typos, incomplete, etc. have made almost all of them fail out of the box. As this is teaching code we should make sure it's perfect :-). See also bug in trac about possible memory leak in r.example.

Non-Technical

  • Discuss role and current composition of GRASS-PSC
  • Discuss GRASS 6.4.3 and 7.0 release schedule

Semi-Technical

  • Improve Web site or use CMS
  • GRASS Book 4th edition in conjuction with GRASS GIS 7.0 release
    • oversee available data and examples
    • ..

Technical

OSGeo4W

Android

  • continue the Android compilation (needs especially the elimination of XDR in rast and rast3d)

QGIS-Sextante-GRASS interface

Update the GDAL-GRASS interface to GRASS 7

Rules Based programming and Agent Based Modelling

  • Tutorial/Documentation on g.infer
  • r.agent in GRASS 7

Time series

  • Temporal GIS framework demo and discussion
  • Temporal GIS wxGUI and WPS integration
  • massive parallel computation of spatio-temporal datasets

Mobile: inventory of mobile and touch implementations

  • Geopaparazzi, v.in.geopaparazzi
  • GRASS on Android (patches by MN)

Cloud computing

  • g.cloud
  • cluster usage best practices

GSoC 2012

  • v.net.* front end, WMS
  • Python API
  • Image Segmentation

GRASS 7

Model integration

GRASS modules fixes

  • i.atcorr debugging
    • Figure out how to get visibility/aerosol depth at 550nm (perhaps from MODIS atmosphere?)
    • Under certain circumstances i.atcorr produces NULL data
  • GRASS 6.4.3 - Release critical issues to be fixed

Signal handling

  • The Python based scripts should catch CTRL-C to avoid that clutter is left behind and that ugly traceback is printed into the terminal.

Suggestion by Anne Ghisla (yet not working... ):

Index: lib/init/grass.py
===================================================================
--- lib/init/grass.py   (revision 51604)
+++ lib/init/grass.py   (working copy)
@@ -30,7 +30,12 @@
 import subprocess
 import re
 import platform
+import signal
 
+# catch user-sent CTRL-C
+signal.signal(signal.SIGINT, lambda a,b:cleanup())
+signal.siginterrupt(signal.SIGINT, False)
+
 # Variables substituted during build process
 if os.environ.has_key('GISBASE'):
     gisbase = os.environ['GISBASE']


  • Currently if vector creation modules fail to create/import they still leave behind the empty skeleton of a map, so you need to use --o for the next time. It would be nice if cleanup could occur. This doesn't happen (visibly) with raster maps as they are composed in $MAPSET/.tmp/ and only moved into the main $MAPSET elements once everything is finished. Vector maps on the otherhand write to the MAPSET in-place.

Fix wxGUI problems

A. Clerici: please fix:

  • ticket https://trac.osgeo.org/grass/ticket/1564
  • ticket https://trac.osgeo.org/grass/ticket/1570
  • wxGUI bugs (GRASS6.4.2RC3 under Fedora 16):
    • In the 'Create histogram of raster map' function, the style pie and the Color for text and axes options are not working. The options run fine on the tcltk GUI and in command line mode.
    • In the 'Add scalebar and north arrow', the scale can be modified only once and the option can't be re-entered nor the scale can't be deleted. Also in this case there are no problems with the tcltk GUI.
    • the option 'Add raster cell arrows' seems not to work properly. It seems that the arrows in the Map Display cannot zoomed in. It can be checked by the following steps:
      • Load and display the aspect (Spearfish) map.
      • Open the d.rast.arrows panel through the GUI button and choose the same map aspect. Click Apply: the Map Display appears green as expected for the very high resolution.
      • Choose 10 for the 'Draw arrow every Nth grid cell:' option and none as 'Color for drawing grid or “none”' option. The arrows are displayed on the cells of aspect map.
      • To make the arrows visible choose slope as 'Raster map containing values used for arrow length:' and 10.0 as 'Scale factor for arrows'. Zoom in a very small area: the cells of aspect map are enlarged, but the density of arrows remains the same
      • The direction is correctly the same for all arrows inside the same cell but it seems that instead of having an arrow each 10 aspect cells, there is an arrow each 10 pixel of the screen (with tcltk GUI the procedure works fine)

Layout

Optionally, have all-in-one GUI window (requested by many new users!):

Current wxGUI layout with detached window components
Proposal for wxGUI layout modification (Recomposition of existing toolbars, mapview and menus by simple rearrangement into one frame)

Proposed solution: wxLayoutAlgorithm(), see WxGUI#What_others_do

r.li.setup

work on r.li.setup for GRASS7

Saving History

  • when using the wxGUI, it needs to write back to the shell history